nonlinear junction detector physics/spectral purity (harmonic radar/bug detector

[CopperCone]

How much energy is reemitted from a 'harmonic radar'? Like, the efficiency of a device being subjected to microwave energy as a frequency multiplier?

You can detect a semiconductor material (vs a metal oxide natural diode forming on a corrosive ) material by comparing the amount of energy remitted as 2nd and 3rd harmonics.

How much energy is this? How spectrally clean does the oscillator circuit need to be so that the energy being picked up by the antenna is >> (significantly greater then) the harmonic energy of the source?


What kind of ratios of 2nd to 3rd harmonic are you looking at?

Let's say you are scanning something under a material with a nonlinear frequency transparency, in this case for the sake of calibration and reflection errors, it would be best to use a super spectrally clean source, so that all significant energy of the 2nd and 3rd harmonics is generated by subject, and not reflections from the subject due to oscillator impurity.

So, a harmonic distortion standard must be established. How efficient is a material at re-radiating the energy it absorbed as higher frequency indicators of its composition?

This way the error term generated by the reflection of the stimulus sources harmonics is negligible... but how many db's down should I be  looking at (in comparison to peak power of the stimulus signal?)

I am not looking for a solution by nulling the measurement due to the unknown behavior of the 'camouflage material' that is in reflection and absorptivity losses. The only acceptable way would be to make a test standard, with different subjects under the same camouflage, so readings may be established. Sufficient spectral purity is a must, but eventually insertion loss becomes significant. I seek to determine how much power I need up front, and how much filtering I need.

[cdev]

Is it normal for say, two strong FM band signals to be mixed together and produce by-products within the HF band due to nonlinerarities?

Would it be possible to locate the source of those nonlinearities that are controllable (maybe rusty old fences?) and eliminate them. (or at least bend them so they no longer touch one another)

Another possibility is dimmer switches connected to house wiring. Is it possible to build some device to suss out these possible RFI sources?

[dmills]

I would note that there is a subtle point to this, the spectral purity requirements of the source depend strongly on the field strength you manage to generate at the target, due to the fact that the target devices harmonic level is more then linearly dependent on the received power.  Turn the drive power up far enough and eventually detectable signal will exceed the level of the harmonic from the source, the math is similar to second or third order intercept.

It is well known in the RF game that bridge rectifier diodes in linear power supplies can be trouble in a strong RF field (They act as a diode mixer and produce both AM and PM components that are very audible as a buzz on the received signal).

73 Dan.

[CopperCone]

@ evb149

point 1
But if you're assuming the RX is nonlinear enough to matter then you can maybe make the source multi-tone so that the IMD of the two "pretty clean" source components produced by mixing inside the RX is sensed and so moving the detected frequency somewhat away from the fundamental frequency of the two source carriers.  Of course then your sources can themselves mix and produce IMD so I guess in a way it is turtles all the way down... but still it might help if you're worried about -1XXdBc @ X Hz offset source purity vs -1XX dBc received signal levels and not noticing the difference between source THD and targed HG.

point2
For some frequencies I suppose you could even try to get fancy with a swept source and looking at the effective FMCW RX signal... might be more trouble / noise in sweeping the source than it is worth though.

point1:
What do you mean that the sources can themselves mix? You mean in the front end path of the instrument being used? A spectrum analyzer designer takes measures to prevent this correct? What measures can be taken to minimize this (i.e. most SA have a built in YIG preselector. Would you recommend adding additional preselection?

and add alot of isolators between the two power amps connected to the same antenna right? to prevent backfeed and mixing in the PAs themselves.
Would tube based equipment reduce the amount of mixing that occurs?

point 2
Can you elaborate how this is different then a single tone? Other then making the apparent bands 'fat' or physically wide on the SA display. Does some sort of 'mixing' still occur in this situation due to the time of light delay in the 'bug under test'? I find it difficult to imagine this 'swept' mode of operation. It seems to me that if you look at any instant (i.e. froze the instrument displays/time), it will be the same as a single tone... but I'm guessing this is not true due to different speed of light at different frequency in the bug/material being tested? But I would expect this to be tiny, unless the sweep is very fast. Am I missing something? Can bias effect the 'mixing' effect (i.e. different amounts of DC voltages with non negligible drain impedances arise due to a sweep (as various parts of the device being probed rectify with various efficiencies?).. this is the only 'strong' effect that I can imagine that might have a life-span long enough to match the rate of common sweep generators... otherwise it seems like you would need band limited noise? but that is more like point 1 then what you mention here. I would love for you to elaborate its tremendously exciting.

@dmills

point 1
I would note that there is a subtle point to this, the spectral purity requirements of the source depend strongly on the field strength you manage to generate at the target, due to the fact that the target devices harmonic level is   more then linearly dependent on the received power.  Turn the drive power up far enough and eventually detectable signal will exceed the level of the harmonic from the source, the math is similar to second or third order intercept.



point 1
Do you have an example of this nonlinearity you are describing? Some rules of thumb for power level (or references to source material?)

I will read up on second / third order intercept in the mean time.

hopefully this will prevent me from pounding the rubble (and my wallet)

[CopperCone]

I believe I understand what you wrote. This is interesting, excellent post.

Also, a helpful feature might be a switched high isolation diplexer and peak hold on the spectrum analyzer for this measurement.. this may decrease the distortion effects caused by the SA significantly more then just the preselector (so long the pass bands and frequencies are space apart properly).

the attenuation you describe seems difficult, I see a 13 stage LPF has about -60db on the 2nd harmonic, so you would need 2-3 of these (and the highpass equivalent). This would cost several thousand dollars but it is not unreachable, a difficult requirement however, and the equivalent highpass and lowpass filters for the PAs (I believe it would offer better isolation then a isolator for the price? perhaps I am mistaken). airman baseball game singer

The dish geometric isolator concept seems difficult to model, but perhaps it is achievable in EM solver software for the experimenter (or can this thing be done on paper?)... absolutely no idea on this aspect. It seems that it would be highly susceptible to the geometry of the target and nearby obstructions. I am not sure about this and I require further thought. do not take my analysis of this geometric isolator with a grain of salt however, my understanding of EM reflections is at this point extremely primitive.

perhaps two narrow band antenna would work ok.. but  directivity is not great (and horns have fairly wide bands). I am not sure what a narrow band directional RF antenna looks like.


I am still reading about FMCW, so bear with my questions for a while.

[CopperCone]

I have read about FM-radar. It seems that the only advertised parameter is ability to measure range 'instantaneously'. It's relevant to bug detection is not apparent to me.

I guess I need to understand this,

is FM modulated harmonic radar for ease of data gathering, where sweep rate is irrelevant, but in the interest of practically (speed of search) you want to see some kind of anomalies of particular frequencies (i.e. resonance behavior of circuit being tested, etc), so if one had enough patience you could 'piece together' this information using superposition of different spectrograms,

or, is it proportional to sweep speed, that is some kind of unique effect develops as the frequency continuously changes along the circuit (I can kind of imagine some low level effects happening, relaxation periods (I guess due to thermal 'nodes' that form)),

I'm not an expert on wave theory but I can imagine that if the signal was swept fast enough, the variable speed of light in the medium being tested would lead to actual mixing products occurring. This is easier for me to imagine with a voltage controlled oscillator that has a analog sweep, not in increments (i.e. not from a DAC), but I am not sure of the sweep speed required for this to actually take place, in my mind the sweep of a typical generator is slow enough that all the energy from one band passed through the sample before the next band energy arrives... but with a voltage controlled oscillator analog system (not a DAC stepper)... the signal is continuously in changing in every instant.. but not by much given the slew rate limitations of most sweepers?

I figure that I can attempt a calculation using the speed of light... but it seems for a continuous system a integral must be used to see what sweep rate is required... but then the frequency dependence of the delay in the sample would need to be known. I can imagine it as a train car that slows down and crashes into itself some how as it enters a tunnel? (then it exits with a chaotic ripple pattern like shooting a shotgun into a pond from a long distance and watching the ripples).

I am not sure how to setup my calculation, total span of the delay line frequency dependence, probability of intersection of the two waves in the sample... and if a relativistic (unrealistic) sweeper is required for this effect to be significant/measurable.

Does anyone with some rule of thumb knowledge of the factors listed here know if setting up the calculation is worth practicality ? (while I would like to develop my own math problem I am looking for theory that applies to practice). It would be a good rule of thumb indicator for a microwave designer to know however.. to kind of understand when some kind of method becomes available with a advance in technology.

It's fascinating.

Can the propagation delay through something like a silicone die be estimated, or would I need to make a radar to actually measure a pings return time (I can't imagine how fast the clock would need to be for this to be measured)...

I guess this effect might be more useful if the sample is cryogenicly frozen? My hunch is that this method would not produce something easy to measure... but you never know!

perhaps a good analogy could be the 'lifetime' of a microwave in a integrated circuit. Is something like a processor significantly different then a carpenters nail?

********
To summarize my overloaded brain, is there anything to the 'sweep' of the harmonic radar other then data gathering ease? Will it have some (measurable) behavior of the two-frequency harmonic radar?

 and what is the maximum sweep rate of a controlled oscillator (currently known)?

also, the point about power line noise and general capacitive/other leakage stemming from the generators, amplifiers and their bodies is very good. it's a real trap for young players potentially. I would have missed it. Recommendation : long low loss coaxial cable runs between test setups. Separate electrical circuits.

How to filter the power line conducting noise? Granted, the use of high pass filters should keep conducted emissions clean.

[Neomys Sapiens]

To summarize my overloaded brain, is there anything to the 'sweep' of the harmonic radar other then data gathering ease? Will it have some (measurable) behavior of the two-frequency harmonic radar?

As far as NLJ detection is concerned, I have been told that sweeping is not necessary for the basic function to occur, but is greatly increasing the chances of detection due to hitting a favourable size/orientation combination of the 'natural antennas' in the sought device and thereby returning a more solid signal.

I think that this is a pretty fascinating technology - are you intending to build such a beast?